Humanist Discussion Group, Vol. 38, No. 263.
Department of Digital Humanities, University of Cologne
Hosted by DH-Cologne
www.dhhumanist.org
Submit to: humanist@dhhumanist.org
[1] From: Tim Smithers <tim.smithers@cantab.net>
Subject: Re: [Humanist] 38.262: AI, poetry and readers (295)
[2] From: William Benzon <bbenzon@mindspring.com>
Subject: Re: [Humanist] 38.262: AI, poetry and readers (22)
[3] From: William Benzon <bbenzon@mindspring.com>
Subject: Re: [Humanist] 38.262: AI, poetry and readers (127)
--[1]------------------------------------------------------------------------
Date: 2024-11-30 21:21:38+00:00
From: Tim Smithers <tim.smithers@cantab.net>
Subject: Re: [Humanist] 38.262: AI, poetry and readers
Hello
We're sensible thinking folk here. No?
So let's not fool ourselves.
We are not all poets, certainly I'm not, and I don't have a
background in, nor the skills of, any kind of literary
appreciation and criticism, but we can, I think, all tell that
Claude 3.5 Sonnet and ChatGPT, and other so called Generative
AI systems like these, do not produce poems, nor parodies of
poems. People do!
These "poems" people are talking about here are all made by a
person, or persons, first preparing the needed carefully
formed prompt, pushing the RETURN key to make the generative
text system do its stuff, then perhaps deciding to push the
"re-do" option, or deciding to re-do or adjust the prompt and
press RETURN again, and, eventually, perhaps, deciding to
take, the generated text and call it a "poem." No "poem"
would happen without this person, or persons, doing all this
thinking, preparing, judging, deciding, and acting.
What I see going on here, and what I think is really going on
here, is a person trying to make "poems" with the aid of some
automatic text generator which uses statistical patterns found
in the texts left over from much human writing, some of it
human poem writing. [Which is not a new idea, or a new way of
making poems and lyrics!]
There is human intelligence here, but no artificial
intelligence. Neither Claude 3.5 Sonnet, nor ChatGPT, know or
understand what poems are, or anything about how poems are
formed, and how and why they work as poems for humans! They
have no mechanisms for this kind of knowing and understanding
and needed reasoning, let alone any way of deciding to start
themselves up one morning to generate a poem for themselves
and other Generative AI systems to share, enjoy, appreciate,
and critically comment upon, and, perhaps, share with us
humans. These machines are, to put it poetically, but also
literally, stochastic parrots.
We may not all be able to write poems, but we should all see
when there are no clothes being worn. Only humans write,
machines [still] only generate text.
To be clear. Words are what we select and put together to say
something. Text is the visual marks left over when we write
our words, and we often preserve and carefully present these
marks so that we may use them to recreate our words, or so
that someone else may do this, using well formed glyphs and
good typographical design to aid the human reading eyes.
Certainly, therefore, these marks are useful to have, but it
takes humans to read them and recreate the words of the person
who wrote them. The text, the marks left by writing, are not
the words of the author. To have the words again they must be
recreated from the marks of the text, and it takes human
intelligence to do this. We often simplify the way we think
about what goes on here, and thus confuse text with words.
And now, it seems, fall into the silliness of thinking
automated text generation is some kind of writing.
To be clear. It takes a person to decide to present some text
as a poem, whether they wrote the words, and are thus left
with the text marks of their words, or got the text generated
by some gee whiz machine. I bet if the generated text didn't
look somehow like a "poem" to this person, they wouldn't
present the text as a "poem." But, looking-like a poem is not
a sufficient, nor even a necessary, condition of being a poem.
Trying to make "poems" using text generators thus easily
results in what I call 'Artificial Flower' "poems" -- texts
which may look like poems to many people, but ain't poems, not
real ones -- and, cannot result in things I would call
'Artificial Light' poems -- marks that are artificially made
somehow to be used to recreate words well chosen to form real
poems, and which may not look like any poems we've seen
before, but, which, upon genuine analysis and interpretation,
can be shown to be, and thus appreciated as, real poems,
perhaps in new and exciting ways.
-- Tim
PS: None of this was made using a text generator, even if it
looks like it was. Writing _good_ nonsense takes real
intelligence. Read the poems of Edward Lear, Lewis Carroll,
Mervyn Peake, and Spike Milligan, for some examples of what I
think are good nonsense. -:)
> On 30 Nov 2024, at 08:38, Humanist <humanist@dhhumanist.org> wrote:
>
>
> Humanist Discussion Group, Vol. 38, No. 262.
> Department of Digital Humanities, University of Cologne
> Hosted by DH-Cologne
> www.dhhumanist.org
> Submit to: humanist@dhhumanist.org
>
>
> [1] From: Fishwick, Paul <Paul.Fishwick@utdallas.edu>
> Subject: Fred The Heretic, Poet (25)
>
> [2] From: scholar-at-large@bell.net <scholar-at-large@bell.net>
> Subject: A. I. Richards (41)
>
> [3] From: James Rovira <jamesrovira@gmail.com>
> Subject: Re: [Humanist] 38.261: AI poet with a mind of its own? (37)
>
> [4] From: Robert A Amsler <robert.amsler@utexas.edu>
> Subject: Re: [Humanist] 38.261: AI poet with a mind of its own? (49)
>
>
> --[1]------------------------------------------------------------------------
> Date: 2024-11-29 20:17:47+00:00
> From: Fishwick, Paul <Paul.Fishwick@utdallas.edu>
> Subject: Fred The Heretic, Poet
>
> Dear Colleagues @ Humanist:
>
> We have worked with Fred Turner, poet, for the past few months, to create an
> OpenAI custom GPT that we call “FredTheHeretic” or FTH, for short. We now have
> two graduate students (Priya and Mihir) working on the project, which we call
> The CyberPoetry project.
>
> You can find FTH by searching through GPTs, or directly here:
>
> https://chatgpt.com/g/g-XmhqgURbv-fredtheheretic-fth
>
> If you have any questions or critiques, we would love to hear them. We will be
> publishing as we move forward.
>
> My only suggestion, emphasized in the help section, is to be precise with your
> prompts. See the example prompts given when you start FTH.
>
>
> Paul Fishwick, Ph.D.
> Distinguished Chair Emeritus of Arts, Humanities, and Technology
> (AHT<https://aht.utdallas.edu/>)
> Former Professor of Computer Science (UF<https://www.cise.ufl.edu/> and
> UTD<https://cs.utdallas.edu/>)
> CEO Metaphorz, LLC (Consulting)
> Email: metaphorz@gmail.com<mailto:metaphorz@gmail.com>
>
> --[2]------------------------------------------------------------------------
> Date: 2024-11-29 19:17:04+00:00
> From: scholar-at-large@bell.net <scholar-at-large@bell.net>
> Subject: A. I. Richards
>
> Willard
>
> This may be of interest to those working in machine learning. It was created
by
> Leonardo Flores
>
> [cite]
>
> Preface to Principles of Cyborg Criticism by A. I. Richards
> This book was inspired by a brief conversation with Ted Underwood on Bluesky
> responding to a study that as Ted summarized it found that “non-expert readers
> prefer AI generated poetry to the classics” and cited famed New Critic I. A.
> Richards and the gulf between expert and non-expert readers. My punny playful
> response “AI Richards?” led him to elaborate with an idea that is at the heart
> of this generated book: “We should create it! A language model that is
convinced
> no human readers really understand poetry.”
>
> [/cite]
>
> https://github.com/NaNoGenMo/2024/issues/27
> <https://github.com/NaNoGenMo/2024/issues/27>
>
> Both the transcript of the dialogue between machine and human and the end
> product are available.
>
> The book begins:
>
> [cite]
> A book is a machine to think with, but in this age of accelerating
collaboration
> between human ingenuity and artificial intelligence, it might better be
likened
> to a hybrid engine—drawing on human creativity to fuel the computational drive
> of a new era of understanding. This book, in particular, aims to chart the
> tangled and treacherous terrain of literary criticism in a time when both
human
> and machine voices clamor for meaning. It is an attempt to re-weave not only
> raveled parts of our civilization but also the threads of human and artificial
> cognition into a fabric resilient enough for the future of literature.
>
> [/cite]
>
> Enjoy
>
> François Lachance
>
> --[3]------------------------------------------------------------------------
> Date: 2024-11-29 16:08:21+00:00
> From: James Rovira <jamesrovira@gmail.com>
> Subject: Re: [Humanist] 38.261: AI poet with a mind of its own?
>
> I think they're good parodies that could generally pass for being written
> by a human being. What makes formal poetry less detectable as AI than
> student papers is that meter and rhyme are predetermined, and when you're
> writing a parody, certain word choices are also predetermined.
>
> I've read plenty of parody poems that read just like those.
>
> Jim R
>
> On Fri, Nov 29, 2024 at 3:44 AM Humanist <humanist@dhhumanist.org> wrote:
>
>>
>> Willard has talked of how AI are minds of their own kind and need to be
>> understood in such terms. This post speaks to that in an oblique way. I
>> asked
>> Claude 3.5 Sonnet to produce three parodies of “Kuala Khan.” The poems it
>> produced are bad poems, which is what I expected. But it seems to me that
>> they
>> are bad in a way that no half-way decent human poet would produce.
>>
>> Bill B
>>
>> https://new-savanna.blogspot.com/2024/11/claude-parodies-kubla-khan-using-
>> elon.html
>> <https://new-savanna.blogspot.com/2024/11/claude-parodies-kubla-khan-using-
> elon.html>
>>
>> --
> Dr. James Rovira <http://www.jamesrovira.com/>
>
> - *David Bowie and Romanticism
> <https://jamesrovira.com/2022/09/02/david-bowie-and-romanticism/>*,
> Palgrave Macmillan, 2022
> - *Women in Rock, Women in Romanticism
> <https://www.routledge.com/Women-in-Rock-Women-in-Romanticism-The-
> Emancipation-of-Female-Will/Rovira/p/book/9781032069845>*,
> Routledge, 2023
>
> --[4]------------------------------------------------------------------------
> Date: 2024-11-29 14:13:33+00:00
> From: Robert A Amsler <robert.amsler@utexas.edu>
> Subject: Re: [Humanist] 38.261: AI poet with a mind of its own?
>
> AI software has no concept of what it is producing. It at present only
> attempts to replicate the grammar and stylistics it encounters in original
> works it analyzes. I suspect the human designers have no concept of the
> intent of the original work. That would require a major effort to create a
> model of what the original work was attempting to express. Such an effort
> could be attempted, but I doubt it could be built to scale. One could
> presumably ask an AI program to explain the reasoning in an original work.
>
> It might be possible to do this for some works--but humans have more
> background knowledge of what has been said about the original text. Asking
> the AI software to interpret what has been said about the original work by
> reading the all mentions of the original work in other texts might come
> close to appearing to express what humans have said about the work. I think
> it would still fail to make new observations. I don't think AI's at present
> will be doing much more than explaining what humans have previously said
> back to us.
>
> On Fri, Nov 29, 2024 at 3:44 AM Humanist <humanist@dhhumanist.org> wrote:
>
>>
>> Humanist Discussion Group, Vol. 38, No. 261.
>> Department of Digital Humanities, University of Cologne
>> Hosted by DH-Cologne
>> www.dhhumanist.org
>> Submit to: humanist@dhhumanist.org
>>
>>
>>
>>
>> Date: 2024-11-28 19:52:41+00:00
>> From: William Benzon <bbenzon@mindspring.com>
>> Subject: NEW SAVANNA: Claude parodies “Kubla Khan” using Elon
>> Musk, FDR, and Walt Disney as subjects
>>
>> Willard has talked of how AI are minds of their own kind and need to be
>> understood in such terms. This post speaks to that in an oblique way. I
>> asked
>> Claude 3.5 Sonnet to produce three parodies of “Kuala Khan.” The poems it
>> produced are bad poems, which is what I expected. But it seems to me that
>> they
>> are bad in a way that no half-way decent human poet would produce.
>>
>> Bill B
>>
>> https://new-savanna.blogspot.com/2024/11/claude-parodies-kubla-khan-using-
>> elon.html
>> <https://new-savanna.blogspot.com/2024/11/claude-parodies-kubla-khan-using-
> elon.html>
--[2]------------------------------------------------------------------------
Date: 2024-11-30 18:43:12+00:00
From: William Benzon <bbenzon@mindspring.com>
Subject: Re: [Humanist] 38.262: AI, poetry and readers
I’ve asked FTH for some “Kuala Khan” parodies. Here’s the link:
https://chatgpt.com/share/674b40ac-fc74-8012-98b5-0645b37eec91
> On Nov 30, 2024, at 2:37 AM, Humanist <humanist@dhhumanist.org> wrote:
>
> Dear Colleagues @ Humanist:
>
> We have worked with Fred Turner, poet, for the past few months, to create an
> OpenAI custom GPT that we call “FredTheHeretic” or FTH, for short. We now have
> two graduate students (Priya and Mihir) working on the project, which we call
> The CyberPoetry project.
>
> You can find FTH by searching through GPTs, or directly here:
>
> https://chatgpt.com/g/g-XmhqgURbv-fredtheheretic-fth
>
> If you have any questions or critiques, we would love to hear them. We will be
> publishing as we move forward.
>
> My only suggestion, emphasized in the help section, is to be precise with your
> prompts. See the example prompts given when you start FTH.
--[3]------------------------------------------------------------------------
Date: 2024-11-30 09:53:22+00:00
From: William Benzon <bbenzon@mindspring.com>
Subject: Re: [Humanist] 38.262: AI, poetry and readers
Continuing on, I offer you “Thirteen Ways to Think about an A.I” – https://new-
savanna.blogspot.com/2024/03/thirteen-ways-to-think-about-ai-3qd.html.
I used the same procedure with the LLM (ChatGPT this time) that I did to get the
“Kuala Khan” parodies. I gave it the original poem and asked it to create a new
on by substituting “A.I.” for “blackbird.”
Note the comment I’ve inserted below.
Bill B
[snip]
>
> --[3]------------------------------------------------------------------------
> Date: 2024-11-29 16:08:21+00:00
> From: James Rovira <jamesrovira@gmail.com>
> Subject: Re: [Humanist] 38.261: AI poet with a mind of its own?
>
> I think they're good parodies that could generally pass for being written
> by a human being. What makes formal poetry less detectable as AI than
> student papers is that meter and rhyme are predetermined, and when you're
> writing a parody, certain word choices are also predetermined.
>
> I've read plenty of parody poems that read just like those.
>
> Jim R
>
> On Fri, Nov 29, 2024 at 3:44 AM Humanist <humanist@dhhumanist.org> wrote:
>
>>
>> Willard has talked of how AI are minds of their own kind and need to be
>> understood in such terms. This post speaks to that in an oblique way. I
>> asked
>> Claude 3.5 Sonnet to produce three parodies of “Kuala Khan.” The poems it
>> produced are bad poems, which is what I expected. But it seems to me that
>> they
>> are bad in a way that no half-way decent human poet would produce.
>>
>> Bill B
>>
>> https://new-savanna.blogspot.com/2024/11/claude-parodies-kubla-khan-using-
>> elon.html
>> <https://new-savanna.blogspot.com/2024/11/claude-parodies-kubla-khan-using-
> elon.html>
>>
>> --
> Dr. James Rovira <http://www.jamesrovira.com/>
>
> - *David Bowie and Romanticism
> <https://jamesrovira.com/2022/09/02/david-bowie-and-romanticism/>*,
> Palgrave Macmillan, 2022
> - *Women in Rock, Women in Romanticism
> <https://www.routledge.com/Women-in-Rock-Women-in-Romanticism-The-
> Emancipation-of-Female-Will/Rovira/p/book/9781032069845>*,
> Routledge, 2023
>
> --[4]------------------------------------------------------------------------
> Date: 2024-11-29 14:13:33+00:00
> From: Robert A Amsler <robert.amsler@utexas.edu>
> Subject: Re: [Humanist] 38.261: AI poet with a mind of its own?
>
> AI software has no concept of what it is producing. It at present only
> attempts to replicate the grammar and stylistics it encounters in original
> works it analyzes. I suspect the human designers have no concept of the
> intent of the original work. That would require a major effort to create a
> model of what the original work was attempting to express. Such an effort
> could be attempted, but I doubt it could be built to scale. One could
> presumably ask an AI program to explain the reasoning in an original work.
That’s certainly possible, though I didn’t attempt it. I’d asked it for another
Musk version, which it ended with the following coupled:
It was a vision of bold device,
A virtual empire built on binary precise!
In then asked it to explain what it was doing. Here’s what it said:
Ah yes! This closing couplet is particularly significant as it directly mirrors
Coleridge's original ending:
It was a miracle of rare device,
A sunny pleasure-dome with caves of ice!
I crafted this parallel deliberately with several layers of meaning:
1. Where Coleridge uses "miracle of rare device" to describe Kubla Khan's
physical pleasure dome, I used "vision of bold device" to play on multiple
meanings of "device":
o The literal electronic devices that form the foundation of Musk's companies
o The literary meaning of "device" as a plan or scheme
o The ambition or "device" of Musk's various ventures
2. The contrast in Coleridge's original between "sunny" and "ice" creates a
paradox that makes the pleasure dome seem magical. I mirrored this with my own
technological paradox:
o "Virtual empire" suggests something intangible
o "Binary precise" grounds it in the exact, mathematical nature of computer
code
o Together they create a similar tension between the ethereal and the concrete
3. The word "empire" is intentionally loaded:
o It references Musk's various companies (Tesla, SpaceX, X/Twitter, etc.) that
together form a kind of technological empire
o It carries some of the same grandiose, perhaps hubristic overtones that
Coleridge was suggesting about Kubla Khan's pleasure dome
o The word "virtual" modifies it to suggest both digital technology and
perhaps a critique of how solid this empire really is
4. "Binary precise" also works on multiple levels:
o Literally refers to computer binary code (1s and 0s)
o Suggests Musk's often black-and-white thinking and pronouncements
o Creates a stark contrast with the more dreamlike imagery earlier in the poem
The exclamation point at the end, like in Coleridge's original, leaves us with a
sense of wonder but also perhaps a note of warning about the grandiosity of the
vision described.
_______________________________________________
Unsubscribe at: http://dhhumanist.org/Restricted
List posts to: humanist@dhhumanist.org
List info and archives at at: http://dhhumanist.org
Listmember interface at: http://dhhumanist.org/Restricted/
Subscribe at: http://dhhumanist.org/membership_form.php
